Tag: pci (Page 1 of 3)

PCI-DSS Card Data Discovery Scans


For PCI-DSS, there are some fairly obvious requirements that are set in stone in order for you to pass PCI-DSS. ASV scans quarterly. Internal vulnerability scans – quarterly. Annual penetration testing. Half yearly reviews of firewall config and policies. Annual training awareness. These are biblical principles of the gospel of PCI.

And then again, there are other areas where interpretation is a little more of a touch and go; up in the air; subjective to the wind; sort of the things where there are as much disagreements and controversies as whether Han shot first or Greedo was just an absolute tool who misses from two feet.

And while most arguments often stems from our clients and us as we try to explain some concepts to them, there comes once in a while a subject where we find ourselves against the explanation of QSAs. Now, not all QSAs are created equal. When I say QSAs here, I refer to the individual QSA, not the organisation QSA. As in the human being who are QSAs for the QSA-C (QSA Company). We’ve worked with some who are technically well versed; we’ve worked with some who are strong in documentation and theory, we’ve worked with some who can communicate well but not so technical, and those who are opposite. But every once in a while, we come across QSAs who think they know everything (they don’t), and they stubbornly stick to a point of argument even when we have exhausted all avenues to show them their point is flawed. The more we argue, the more adamant they take their stance even if their justifications seem to be plucked directly out of their …. posterior appendages.

One of the items you will often see coming up in PCI-DSS is this thing called the Credit Card Discovery Scanner (CDD). What is this? In PCI-DSS standard pg 10:

To confirm the accuracy of the defined CDE, perform the following:
The assessed entity identifies and documents the existence of all cardholder data in their environment, to verify that no cardholder data exists outside of the currently defined CDE.

PCI DSS v3.2.1

The CDD process is basically just a process using a tool usually to identify whether card information is stored in the clear within the organisation. These are usually regular expressions based applications; where it can categorise the type of card based on BIN or the initial numbers. These tools are often quite useful as well to find other forms of information like personal information etc, as long as you can identify filters and regular expressions for them. Some tools out there are from Groundlabs, Managed Engine, ControlCase etc. We also have free CDD tools like Pan Buster, Credit Card Scanner etc. The free tools are a little bit more difficult to use in our opinion and there seems to be less support for database scans and more false positives overall, so you may spend a longer time cleaning up the results.

Whether commercial or free tools, what PCI has been fairly silent about is whether these are mandated in the standard to be done. Unlike ASV scans or penetration testing, the standard doesn’t specifically state the need to run these tools for a normal PCI-DSS standard. When I say ‘normal’; I refer to a set of additional requirements under Appendix A3: Designated Entities Supplemental Validation (DESV) . These are specially assigned entities that has large volume of card data or has suffered significant breaches. This is designated by payment brands or acquirers, and it’s not something a QSA or even the audited entity decides on.

So looking into the card data scan requirements; we only have the Pg 10 scoping requirement and in the DESV portion , A.3.2.5 – “Implement a data-discovery methodology to confirm PCI DSS scope and to locate all sources and locations of clear-text PAN at least quarterly and upon significant changes to the cardholder environment or processes”

In most cases, CDD scans are done on an annual basis for normal PCI-DSS (non DESV), or at times half-yearly as required by the QSA.

So along came another QSA who stoutly declares that all companies are required to do a quarterly CDD scan regardless of size for all systems in scope. When politely reminded that he seems to be mixing up the DESV quarterly scan requirements; he says no. He is highlighting requirement 3.1: “A quarterly process for identifying and securely deleting stored cardholder data that exceeds defined retention.”

When pressed to explain why this is a CDD scan, he states its obvious, that everyone needs to run the CDD scanner every quarter to address this requirement.

OK. We disagree. Completely. This is one of the instance, where QSA super-imposes requirements on each other just because it sounds the same.

Let’s break it down by looking at the PURPOSE of the CDD scan. And the best way is to go back to the standard and pick up the part where the standard states a ‘data-discovery’ method in DESV A3.2.5.

Implement a data-discovery methodology to confirm PCI DSS scope and to locate all sources and locations of clear-text PAN

A3.2.5 PCI-DSS V3.2.1

It’s clear that the CDD purpose is to locate where CLEAR-TEXT PAN is found in the CDE (and non-CDE) environment. Why is this important? Because in the CDE, there should never be any clear-text PAN found in storage. All PANs must be protected by either of the Four Horsemen of the Apocalypse: Encryption, Truncation, Hashing or Tokenization. A failed CDD means there are card PAN found in clear text within the CDE.

So with that in mind, lets go back to requirement 3.1. This is nothing to do with identifying clear PAN. It talks about identifying AND deleting EXPIRED card data (based on retention policies). That’s it. If the PAN is encrypted or tokenized but its stored beyond its retention period; requirement 3.1 tells you to delete it. It talks about retention period and storage beyond it. Which part of it talks about doing a card data scan to identify clear text card information?

In the description, it further states: A quarterly process for identifying and securely deleting stored cardholder data that exceeds defined retention requirements.

So QSA, please RTFM; requirement 3.1 isn’t talking about the need to run CDD quarterly to identify clear-text PAN storage; it is to run something (script) or manual; to identify PAN storage that is already expired. It is to discover duration of storage; not security of storage. Running a shell script may be good enough to get the timestamp of files; or checking the timestamp on the database entries to ensure that all card data is removed or anonymized after a period of say, 7 years.

If you need assistance in PCI-DSS or any other compliance standards like the ISMS or ITSM, drop us a note at pcidss@pkfmalaysia.com. We can help clarify some of these annoying requirements that even QSAs (as experienced as they are) are plucking out of their rear appendages.

PCI-DSS 2022 and Version 4


So we are now in 2022. PCI-DSS v4.0 is due to be out and one of the things we have been doing for the first two weeks of the year is to get over our holiday hangovers. That’s right. In our country (Malaysia), the slowest months are December, January and February. It’s like starting a car in the dead of winter. These 3 months are like the Amen Corner in Augusta for businesses. December hits like a ton of bricks due to the Christmas season; and then just when things start moving in January, it grinds to a halt for Chinese New Year, where the entire nation just flat out refuses to work. When we are back in the second week of Chinese New Year, we are once more in first gear climbing up the hill again of 2022.

So we did things a bit differently. We started the first two weeks with a series of training for clients and potential clients, to go through PCI-DSS v4.0 and create an awareness of what is there to expect.

The above is taken from the PCI website and immediately we see some interesting things here. Number one: PCI-DSS v3.2.1 only retires in 2024. This is interesting, because usually the transition period isn’t so long. It’s long now because – I don’t know, there may be an ongoing pandemic and such. So here we are Q1 2022, and our customers are asking when do we transition to v4.0?

Well, the answer would be: as soon as you can. But in theory, you can probably stick to v3.2.1 validation for 2022 and realistically move to v4.0 in 2023. In fact, for some of our clients whose PCI maintenance period follows the calendar year, they can even force 3.2.1 into their 2023 validation year.

As for the actual content in PCI v4, it’s still a well kept secret like the plot of Spiderman: No Way Home; but we have been reading a bit and also have joined last year’s PCI-DSS community meeting and learnt some interesting tid-bits of it.

No 1: Compensating Controls

The-get-out-of-jail-free card. Customers have been dangling this Compensating Controls card in front of our faces ever since the Mesopotamian times. When they can’t address a control – use compensating controls! When they cannot implement something due to budget – compensating controls! When they can’t make changes to an application because it was designed by a group of kindergarten kids and it would break the moment you touch it – Compensating Controls! When you don’t know what to say to your wife after a long night out at the pub with the mates and come back smelling like a keg of kerosene – Compensating Controls!

The problem with compensating controls is that they are a pain in the neck to implement and to document. And to justify. The compensating control worksheet, the justification documentation, the implementation of the control itself to be ‘above and beyond’ the scope of PCI-DSS etc. Everyone things this is a silver bullet only to find it the deepest rabbit hole you can ever fall into.

So, PCI v4 does away with compensating controls. Great.

And they introduce Customized Implementation.

A lot of people are saying this is a game changer.

Honestly? Until more information comes out, we only have this to go with:

Customized implementation considers the intent of the objective and allows entities to design their own security controls to meet it. Once an organization determines the security control for a given objective, it must provide full documentation to enable their Qualified Security Auditor (QSA) to make a final decision on the effectiveness of a control.


Design their own security controls? Well, ok, isn’t this the same as compensating controls? I am thinking this just expands the interpretation to something a bit broader in which case the control may not even be a technical control. So instead of stating , ok, we can’t meet certain password controls due to the legacy application issue, and compensating controls were previously excessive logging and monitoring; isolation of network, whitelisting of IPs and access; using WAF and DLP and Virtual patching etc etc; are we stating now, a possible customized approach would be: instead of all these technical controls; we now have a customized security approach. Which includes isolation of network, whitelisting of IPs and access; using WAF and DLP and Virtual patching etc etc.

Until we see some examples of this, it may just be well that most companies will go along with the ‘normal’ approach; or adopt a wait and see approach and eke out the last remaining drop of v3.2.1. into 2024.

No 2: UP in the Clouds

Another item that has been long overdue? Cloud. It’s about time things get addressed and not just cloud, but how services and containers work as well. We have had auditors coming to our clients insisting on them doing testing, VA/PT on services from AWS, not recognizing there’s not even an IP address to start with. To be fair to the SSC, they do have a few Cloud Guidelines Supplementary documentation, which we actually find very useful especially in our projects on certifying cloud technologies. We can see this being incorporated more formally into v4.0 where the requirements will be designed around Cloud environment more organically than what we see right now (sort of force-fitting many of the traditional concepts like Network IDS, Patching etc into the cloud environment).

No 3: Not another MF-A!

I have a bad feeling about this.

MFA has been a constant pain for us. Firstly, where MFA is being implemented – not just on perimeter but now on every access to the CDE. At least it’s now still on admin accounts. We hear they plan to introduce for ALL users. We also hear the collective screams of the tormented from the nine hells of Dante. Secondly, a lot of customers are still depending on MFA via SMS. If PCI goes along the NIST route, we could see this being deprecated soon. Also, clarifications as well on whether client side certificate are acceptable as a ‘something you have’ factor would be most welcomed. We see different QSAs interpreting this so differently you’d think we’ve asked them to interpret some ancient Thuggee text. Multi-factor challenges are already there for us over the past years, with Bank Negara’s RMIT focus on ‘strong MFA’ for large financial institutions. A clear guidance also should be there on how to evaluate multi-factor that is dependent on a cloud provider; and whether common implementation like Google Authentication etc can still be considered as good enough for V4.0

No 4: Encrypting everything

We also hear now that the “Pocket Protector Trope” security may be implemented. Remember those movies we watch, where the hero gets shot in the chest and you think he dies but he reveals that the bullet is stopped by his pocket watch; his badge; a bible; or some other dang sentimental thing that was given to him like 40 scenes ago?

So in PCI, usually when data is traversing the internet or network, it states the transmission needs to be encrypted. It doesn’t technically state anything about encrypting the data package itself while in transmission. The data encryption almost exclusive occurs during data at rest. So in this case, they are doubling the protection: They are adding that pocket watch to catch the bullet; so if the transmission gets compromised, the data is still secured. The bullet doesn’t hit the hero!

No 5: Recovery and Continuity

Not so much as something coming, but more of what we’d like to see. One of the biggest criticism we see customers bemoaning at PCI (other than the cost and budget and the complexity and..ok, everything else) – is that PCI has little focus on business continuity and disaster recovery. It’s almost as if PCI is standing there saying, “OK, you have outage for a few days? Great, make sure your credit card information is safe.” It’s not really business focused, it’s more credit card confidentiality focus. What we would like to see is a little more focus on this area. Over the past 2 years, we have seen customers getting all sorts of attacks from cyberspace. Malware, ransomware, hacking, fraud, defacement — it’s like the world goes into a pandemic and everyone’s bored to bits at home and everyone is taking up hacking as a part time gig. Malware for instance – how prepared is a PCI compliant company against ransomware attack? Have they done their backups? Have they tested their systems to recover?

So, if you have any queries on PCIv4 for us, drop us an email at pcidss@pkfmalaysia.com and we will definitely get back to you. Have a great and safe year ahead for 2022!

PCI-DSS and Card Storage


We had an interesting discussion a few weeks back about storage in PCI-DSS. We disagreed with an acquirer’s position in how PCI-DSS views storage and therefore opened a whole can of … interesting debate.

The problem the acquirer had with our position was simple. We have a client who is currently doing a data migration import from another service provider to their document management system. Amongst the terabytes of data were possible scanned copies of credit card information, either in forms or actual card photo-copies themselves. Now, we are talking about terabytes.

Our position was fairly straightforward. Do you need these card data? We asked. No, said our client. We don’t need the card data as we do recon and backoffice operations on other form of identification. Can these information be removed or redacted? Bemused, they said, possibly, but the problem is that there are going to be millions of records to be dealt with.

Well, is there a way we can sanitize the data before it enters into your environment?

Yes, possibly, we need to ask the acquirer to ask their current provider to do it for us.

The provider you are taking business away from?


Good luck…

And sure enough, the acquirer responded and asked us, “Shouldn’t PCI-DSS allow the storage of these card information, and how your client is able to deal with it? Why do you insist on us redacting and removing the card information? What then is the purpose of PCI-DSS??”

Now, on the surface, that argument does make sense. After all PCI-DSS applies to entities who store, transmit and process credit card information right? Why then wouldn’t we want our client to store credit card information if they are going through PCI-DSS?

Unfortunately, this is a case of getting the solution (PCI-DSS) mixed up with the problem(storing card data). In other words, in a more current analogy, just because I got vaccinated doesn’t mean I would purposely go out and try to get infected so that the vaccine has something to do. The purpose of PCI isn’t for you to store credit card. It’s for you to manage the storage of credit card IF you store it. Storing credit card isn’t a PCI-DSS objective, its an issue that PCI-DSS tries to solve.

So back to this little kerfuffle; if they pass us terabytes of information with card data, our client will need to figure a way to protect this data. Likely encryption of any information that card data is present, which includes key management etc. If they can redact it and remove it before it enters into our client’s environment, then we avoid it. We are basically following the concept of PCI-DSS :

Requirement 3 addresses protection of stored cardholder data. Merchants who do not store any cardholder data automatically provide stronger protection by having eliminated a key target for data thieves. Remember if you don’t need it, don’t store it!

PCI-DSS Prioritized approach

If we don’t need it, don’t store it. In this case, we don’t need it, so we are trying to escape storing it. However, if this cannot be done (which likely it won’t be), then we just need to put controls in there. We’re trying to get our clients to do less and we are also trying to remove card footprints in other areas, thus reducing the risks to the card brands, and likely save the world from impending disaster and destruction.

However, we do have another issue.

Because there is potentially CVV storage (photocopy of cards front and back) and scanned into softcopies, we have a bit of a problem. CVV cannot be stored in any format or in any media post authorisation. So therefore, if this is being dumped into our client’s environment, it’s imperative someone removes this information. To us, its a lot easier to remove it at source; but unfortunately that means there is an effort to be spent on it, which no one is willing to do.

How the CVV got stored in the first place is a question that we don’t have an answer to. However, we do know that if CVV is present, we cannot just encrypt it and be done with it. We will need to remove these information one by one. There are a few solutions out there that can do auto redaction and be applied to a massive amount of files, provided that the files are in a sort of standard fashion. That could be a solution on this, but again, it’s beyond what we are discussing for this article.

The point is, having PCI-DSS doesn’t automatically mean we MUST store card data. It simply means IF we store card data we are applying PCI-DSS controls to that storage of card data.

Let us know if you need more information about PCI-DSS or any IT standard compliance like ISO27001 or CSA/SOC, we are ready to assist, just contact us here. Stay safe everyone!

ASV Scans /= PCI Compliance

There is an old story about a chicken and eagle. I hear this story being told by life coaches or motivational trainers trying to get through to our thick, jaded, technical skull that there is something more to life than coding and technology.

The abbreviated version is this: A farmer was walking and finds an eagle’s egg fallen out of the nest. He picks it up, brings it back to his farm, and puts it into the chicken coop. Soon, it hatches, and joins the other chickens in the farm and learns how to be a chicken, even though its an eagle. So this is where some of the version diverges.

a) The chicken and the eagle starts talking one day and the eagle notices another eagle flying high in the sky and he goes, “Dang, I wish I could be an eagle,” and his chicken-pal looks at him scornfully and says, “You are a chicken. How can you be like the king of all birds, soaring through the sky?” So the eagle keeps thinking he is a chicken and the next day he gets roasted for dinner. And the farmer finds his meat a bit tough and doesn’t taste like chicken at all. The moral here is: Don’t let your limitations inhibit you or you will end up a cooked and eaten. This is probably the original version before the other two came along below:

b) The farmer is visited by a naturalist who observes this ‘chicken’ and immediately knows he is an eagle. So he takes this chicken up to a high cliff, and throws him over, shouting: “Spread your wings and fly! Soar like the eagle you are meant to be!” And the eagle soars through the clouds and sky and become the king of all birds. The moral of the story: All of us are eagles, even if you think you are a chicken. All you need is a life coach or a motivational trainer to throw you off the ledge and you will soar. This is the preferred version for life coaches and motivational speakers. For obvious reason.

c) Same as story b) above, but instead of soaring, the naturalist throws the ‘chicken’ off the ledge, and it falls 100 feet and splatters its brains all over the bottom of the ledge and dies since it doesn’t know how to fly. And gets cooked and roasted for dinner. The moral of the story (and this is by far, our more preferred, realistic and risk-averse version): Don’t do something you may be destined for but not ready for. Or you will end up smashed, cooked and eaten.

All three versions have this theme in common: The eagle isn’t a chicken and the chicken isn’t an eagle. The chicken may have commonalities of an eagle, like wings and a beak, but just because it has those doesn’t make it an eagle.

Yes, I am aware that the anecdote above isn’t a very good illustration of the point I am trying to make, but I couldn’t think of a better one. And in a roundabout way, what I want to illustrate here is that ASV scans do not make you PCI Compliant.

We get this a lot.

A company would come and say they are PCI-compliant. Or we have a client who outsources certain portion of their operations to another company and that company comes back and shows us their ASV compliant scan and says this is all they need to show us. We (The auditors/consultants) are compelled to accept this because the ASV scans demonstrate their PCI Compliance, they say.

Let’s make a point here: ASV questions and subquestions in the SAQ D covers around 14 queries. Out of around 600. That means ASV covers 2.33% of PCI-DSS. There is a massive load of other controls and items covering PCI-DSS Other than those precious ASV quarterly scans. What about your patching? Hardening? Firewall security? HR policies? Logging and monitoring? Logical access? MFA? Hardening of systems? Anti-virus and host firewalls? What about service provider management? What about vendor default passwords? What about storage, encryption, key management? Software development? Application and penetration testing? Internal vulnerability scans? Training?

You can see how impossible it is to accept just the ASV report as an evidence of PCI compliance, much like how we cannot accept the chicken as an eagle, but yet, we are constantly berated upon that we don’t know what we are doing and that their Banks have accepted their ASV scans as a sign of PCI compliance, so we should to. But we can’t. We can’t accept 2.33% as a 100% of something. It’s simply mathematically not possible.

So there you go – banks. Why do banks perpetuate this myth that PCI compliance = ASV scans? Why? It’s 2.33% of PCI-DSS! You can’t accept something as an eagle just because it has wings and a beak! There’s really no argument about it.

Here is what 2.3% feels like:

a) The number of Jazz music of all US Music sales in 2013

b) Increase in slot machine spending in New Zealand in 2018 Q1

c) Auto parts industry against the US GDP in 2013

d) Android 6.0 Marshmallow installation for all Android devices in July 2016

e) Thats lesser than the % of freshwater we have on this planet (2.5% of water on the planet is freshwater)

I am sure there’s a lot of 2.33% out there on this planet, but the point we are making is this: It’s not compliance. It’s a small but important part of compliance but it’s not compliance. So no matter what your banks tell you, we can never accept the ASV scan as a sign of PCI compliance. It can be accepted as one of the evidences of PCI compliance amongst many, but not as an evidence of complete compliance.

Now, stop calling a chicken an eagle. Let us know about your questions for PCI or any compliance at pcidss@pkfmalaysia.com.

SAQ A and A-EP, the eternal struggle

Another week, another lockdown struggle, another political instability and another question on the eternal confusion called the SAQ A and A-EP. And this time, it wasn’t so much of us trying to clarify with the customer on this – but us trying to explain to QSAs on it. It just shows how much confusion there is to this thing even after all these years, that even auditors, whose bread and butter is literally on PCI-DSS still struggle to understand it. I don’t blame them – it’s the way that the SAQs are worded, and the confusion that surrounds it that makes it so frustratingly difficult to interpret.

SAQ A by far gets the most mileage. Not because a lot of people are eligible for it, but because at 20+ questions, it’s by default the go-to SAQ for most organisations, whether they are eligible for it or not. I mean, comparing the SAQ D and the A is like scaling Everest vs the little sand hill that your 5 year old kid just built on the beach. Something like that. So everyone (even non-eligible Service Providers) always default to the Open Shortest Path First, which is the SAQ A when they need to choose an SAQ to be “PCI-Compliant”.

However, SAQ A is notoriously difficult to be eligible for and today we are going to look at it. Again. We have often seen everyone anything from storing card information, to hardcopy storage of insurance policies, to doing outsourced call center picketing in front of our office shouting for their SAQ A rights. I mean, let’s start here with SAQ A and A-EP and the difference.

We are not going to focus on the controls in these SAQ, but rather the ‘eligibility’ of it, meaning, on Page 4 of both SAQ under “Before You Begin”. Instead of just repeating all that is typed in there in this article, I will assume those reading this article is keen for a deeper dive into the murky waters of SAQ and not here for a shallow wade – so I am going to assume you have those SAQs right in front of you and I don’t have to delve into the history much, ok?

SAQ A’s story starts off by stating there are TWO types of business who are eligible for it.

a) E-Commerce Merchants

b) MOTO (mail order/telephone order) – card not present

c) Of course, those who do not STORE, PROCESS or TRANSMIT card holder data in ANY electronic format on their system and premise.

Let’s start with MOTO first, because this often confuses people. Straight away those doing MOTO will dance a jig in front of me and gleefully points out that they deserve the SAQ A. All your base are belong to SAQ A, if those nerds like me would understand. Because I usually move them over to SAQ C or C-VT depending on how their call center/MOTO transactions are set up (even B-IP may apply e.g MOTO function on terminal, but mostly MOTO ends up being in SAQ D because they often store card data on file).

Hold on there though. Eligibility of MOTO is tied to the eligibility of c) – i.e you do not store (OK), process (erm, yeah ok, sometimes) or TRANSMIT card holder. Often the transmit and process part is done when you have people on your premise doing MOTO. The moment a phone call comes in – BAM! you are hit. You are done for.

So the ONLY time MOTO is eligible for SAQ A is later described in the SAQ:

Mail order/telephone order (MOTO) or e-commerce merchants that have completely outsourced all operations (where there is no redirection mechanism from the merchant to the third party) and therefore do not have any systems in scope for this SAQ, would consider these requirements to be “not applicable.”


The above is talking about how we can mark Requirement 2,6 and 8 as Non applicable. But notice where it states: COMPLETELY OUTSOURCED ALL OPERATIONS. This means, your company’s MOTO transaction is never done by your own company or on your own premise or by your people or using your technology. You have Completely, irrevocably, irreversibly outsourced the entire function to someone else who is PCI-DSS compliant. Then OK, you cool.

So now we know how to deal with that MOTO part. Oh wait, wait. There is a scenario from one client, where customers actually come over to the counter and try to make payment. However, because they have upgraded everything, instead of dipping or waving that card into a terminal for a POS payment, the counter person whips up a high tech iPad, connects to the companies website, looks at the credit card (while the customer is in front of them) and type out the transaction onto the e-commerce site itself for the transaction. How do we deal with this?

Well. This certainly doesn’t qualify for SAQ A in a strict sense, since this is considered a ‘face-to-face’ channel. However, logically, that transaction is made as an e-commerce, card non present transaction, because the CVV is entered as well and on the merchant end, it qualifies as a e-commerce transaction based on the flow and the fee they are paying. This is an interesting scenario as I would likely look at it as an e-commerce flow, since technically, the customer can do it themselves, but its just that for some reason, maybe they don’t know how, or they can’t figure it out, they go over to the counter to do it. The acquirer certainly doesn’t know about it. But because the information is going through the company’s asset, the company’s line, the company’s network, there would be additional risk they need to consider. In the end, it would be the call of the QSA on how they view this, however, I don’t think this could qualify for an SAQ A channel. It could be technically treated as a SAQ B-IP as we can assume its a terminal, but most auditors, to err on the side of caution may just opt for the full SAQ D on this.

OK, MOTO done.

Now for the e-commerce. I am not going to repeat what I’ve written some years back: https://www.pkfavantedge.com/it-audit/pci-dss-saq-a-and-saq-a-ep-differences-in-a-nutshell/

But I am just going to dive right in where the confusion begins. SAQ A-EP is written in a way that confuses people, and requires some sort of Indiana Jones exploration to figure out what in tarnation they are trying to get at.

So, under Before you begin, the second eligibility point (we call this ITEM 2):

Item 2: “All processing of cardholder data, with the exception of the payment page, is entirely outsourced to a PCI DSS validated third-party payment processor;”

This is confusing. They say – “All processing of cardholder data EXCEPT the payment page”. This means, the payment page actually SITS with the merchant, while everything else is outsourced to PCI third party. This means, this SAQ is eligible for merchants with PAYMENT PAGE (where credit card is entered) residing in their premise. So therefore, if the PAYMENT PAGE is also outsourced, immediately, this SAQ is no longer eligible. In a simple logic:

if (paymentpage) then { read next line;} elseif (!paymentpage) { exit();}

That means, SAQ A-EP doesn’t apply anymore to us if we have outsourced the payment page because this condition is not met, and therefore the if statement should exit, or if you are in a loop, it should end. SAQ ENDS.

The problem is auditors are generally non-programmers and even when the condition is no longer eligible, they keep going!!

And it’s really, the next line that is the confusion of all confusion:

Item 3: “Your e-commerce website does not receive cardholder data but controls how consumers, or their cardholder data, are redirected to a PCI DSS validated third-party payment processor;”

I mean, if we had exited the SAQ loop on the second condition, we won’t need to deal with this nonsense. So let’s break it down. YES, my e-commerce website does not receive card holder data, since I outsourced ALL MY PAYMENT page already to third party. But wait, you are saying “CONTROLS’ how consumers or data are ‘redirected’ to a third party? What? Obviously there is an element of control here, so how do we define ‘control’? Isn’t redirecting to an outsource payment page CONTROL?

The next confusion is the next line:

Item 4: “If merchant website is hosted by a third-party provider, the provider is validated to all applicable PCI DSS requirements (e.g., including PCI DSS Appendix A if the provider is a shared hosting provider);”

Hold up – didn’t we already agree that if the merchant entire website is hosted by a third party PCI provider, this would already not be in SAQ A-EP (see the exit rule of item 2). In fact, isn’t completely outsourcing the website the whole point of SAQ A? What sort of black magic is this?

Item 5: “Each element of the payment page(s) delivered to the consumer’s browser originates from either the merchant’s website OR a PCI DSS compliant service provider(s);”

Look at this wording. Look at it. Tell me that this is not contradicting item 2, the ‘with exception of the payment page’ condition. Let me rephrase item 2:

“You can go for SAQ A-EP if you host your payment page and have outsourced your processing to a PCI third party” –therefore implying that if you don’t host payment page and outsource everything, then another SAQ (SAQ A) applies.

Item 5 slaps Item 2 in the face and goes, “No. SAQ A-EP for you if you host the payment page, or the payment page is hosted by your PCI-DSS service provider. So no, Item 2, you wrong. You dead wrong.”

That usage of the word “OR” in that sentence confuses programmers or those with IT background, I think. This is a logical connector where if condition A OR condition B, if any of this is TRUE or both TRUE, we enter into the loop. Compared with the AND connector, where both needs to be true, otherwise we don’t process the loop. So the above statement is stating “ANY CONDITION WHATSOEVER” since it uses “OR”, will need to apply SAQ A-EP.

In fact, if they had clarified if all of these conditions are connected to each other either through the AND or OR operator, it would makes much more sense to us. It’s like the question, “Are you going to do it now OR do it later?” and we answer “Yes!” because we are indeed doing it now or later, and the question didn’t specify which condition as long as we are doing it.

Anyway, back to the story. The note in SAQ A-EP states:

For the purposes of this SAQ, PCI DSS requirements that refer to the “cardholder data environment” are applicable to the merchant website(s). This is because the merchant website directly impacts how the payment card data is transmitted, even though the website itself does not receive cardholder data.


It is very ominous. It states, even if your website does not receive card holder data, you still impact or ‘control’ how the payment card is transmitted.

All is not lost though, because if you flip back to SAQ, under the SAQ A notes:

For this SAQ, PCI DSS Requirements that address the protection of computer systems (for example, Requirements 2, 6, and 8) apply to e-commerce merchants that redirect customers from their website to a third party for payment processing, and specifically to the merchant web server upon which the redirection mechanism is located


I mean, I don’t know how clear it needs to be. It states in SAQ A “FOR THIS SAQ” – apply to merchants that ‘REDIRECT’ customers FROM their website (merchant website) to 3rd party for payment processing and specifically TO the merchant web server where the redirection occurs.

I am going to clarify the phrase that is underlined. the word “TO” is a preposition of the verb “apply to” in the earlier sentence, i.e this applies to merchants, specifically to their web server etc etc. Why its confusing here is because some may read it as a preposition to indicate direction , i.e redirect customers from their website to a 3rd party, specifically TO a merchant web server etc etc, which basically indicates the redirect is going into a loop (from merchant site to third party back to the merchant site) which doesn’t make sense.

I just want to point this out as I may not be the only one confused with this play of words and irresponsible usage of the preposition “TO”. Only me? Ok, fine.

Anyway – long story short, we used the notes in SAQ A to get out of jail for our client, and the QSA seemed to be resigned to that, noting there is a huge huge confusion with how A-EP is written. You do need to know, A-EP was born after A, so definitely, there would be some contradiction since they weren’t written together. SAQ A-EP is like the grumpy uncle that sits in the corner in your Christmas party and snaps at you when you ask him how he’s doing, while SAQ A is like the uncle with all the presents and all the children running around him and laughing with him as he tells a joke. Ah, SAQ A, we like you a lot.

Anyway – a final note on us, while we can state on PCI side, a full outsourcing of e-commerce payment page to third party qualifies for SAQ A, you do need to think – SAQ A-EP makes sense. The page doing the re-direct could be attacked and compromise and the redirect sent to another ‘payment page’ that looks exactly the same as the actual one. So while you are laughing with SAQ A, you need to take into account not to ignore the reasonable requirements that A-EP puts to you – vulnerability scan, firewall rules, penetration testing – i.e these are all best practice baselines that should be practiced regardless of compliance conditions etc. I would recommend a middle ground and take up a risk approach to it – implement these controls based on a good risk assessment and not just ignore the poor, grumpy SAQ A-EP uncle sitting in the corner. Because he may have a point in terms of security, after all.

Let us know about your experience or questions on PCI, SAQs or any other compliance questions at avantedge@pkfmalaysia.com!

« Older posts

© 2022 PKF AvantEdge

Up ↑